I. Introduction

Currently, Internet of Things (IoT) technology is rapidly developing, bringing significant benefits to our lives in a wide range of fields, from smart homes to smart factories, healthcare, and smart buildings. IoT devices offer significant benefits, but they also pose serious security risks (such as privacy violations, unauthorized access, and DDoS attacks). Network administrators are required to implement countermeasures, such as network segmentation and/or isolation, to address these threats. However, the proliferation of IoT devices makes it difficult for administrators to accurately ascertain the number and types of devices present on their networks, hindering the implementation of effective security measures. To resolve this issue, the establishment of automatic identification technology using device traffic information is urgently needed. Various approaches have been proposed for IoT device identification, including rule-based methods and those utilizing machine learning. Machine learning-based methods, in particular, have demonstrated the potential for high identification accuracy.